Re: [PATCH] mm: memory-failure: use rcu lock instead of tasklist_lock when collect_procs()

[Tong Tiangen]

在 2023/8/21 12:34, Matthew Wilcox 写道:
> On Mon, Aug 21, 2023 at 10:25:34AM +0800, Tong Tiangen wrote:
> > +++ b/mm/memory-failure.c
> > @@ -546,24 +546,26 @@ static void kill_procs(struct list_head *to_kill, int forcekill, bool fail,
> >    * Find a dedicated thread which is supposed to handle SIGBUS(BUS_MCEERR_AO)
> >    * on behalf of the thread group. Return task_struct of the (first found)
> >    * dedicated thread if found, and return NULL otherwise.
> > - *
> > - * We already hold read_lock(&tasklist_lock) in the caller, so we don't
> > - * have to call rcu_read_lock/unlock() in this function.
> >    */
> >   static struct task_struct *find_early_kill_thread(struct task_struct *tsk)
> >   {
> >   	struct task_struct *t;
> >   
> > +	rcu_read_lock();
> >   	for_each_thread(tsk, t) {
> >   		if (t->flags & PF_MCE_PROCESS) {
> >   			if (t->flags & PF_MCE_EARLY)
> > -				return t;
> > +				goto found;
> >   		} else {
> >   			if (sysctl_memory_failure_early_kill)
> > -				return t;
> > +				goto found;
> >   		}
> >   	}
> > -	return NULL;
> > +
> > +	t = NULL;
> > +found:
> > +	rcu_read_unlock();
> > +	return t;
> >   }
>
> I don't understand why you need to modify find_early_kill_thread() at
> all.  It's still true that the caller holds _a_ lock protecting it; the
> comment needs to be updated to reflect that it might be the RCU lock
> or the tasklist_lock (or did you change all callers?), but there's no
> need for this function to take the RCU lock itself, afaics?
>
> .

I've checked that all the paths that call find_early_kill_thread() 
already hold the rcu lock, and there's really no need to hold the rcu 
lock here.
In the next patch version, here only the comments are modified.

Thanks,
Tong.